Lecture 33
Growth and Differentiation of Planet Earth
– Formation of the Core and Moon
Reading this week: White Ch 11 (sections 11.5 – end)
Today – Guest Lecturer, Greg Ravizza
1. Core Formation imprint on the mantle
2. origin of the moon
3. Earth Accretion summary
GG325 L33, F2013
Core Formation continued… Chemical indicators from the Mantle
Recalling the discussion at the end of last lecture…
Two things are apparent:
a) Sideophile elements are not as low in the mantle as would be expected
from pure metal-silicate equilibration.
•
They are 5-350 times more enriched than expected for complete
equilibrium between silicate and molten iron.
•
Volatile siderophiles appear to be even more enriched than nonvolatile ones.
These facts could argue for incomplete equilibration (e.g., a kinetic or
spatial effect), an impure Fe phase (e.g., a chemical effect) and/or addition
of a volatile rich component after core formation.
b) Chalcophile elements are depleted in the silicate Earth relative to
chondrites, but not as depleted as many of the siderophiles are.
This could argue against much S in the core.
GG325 L33, F2013
1
Element abundances of the Earth’s mantle normalised to CI chondrite and Ti (data from Palme and O’Neill,
2003). Siderophile elements have metal–silicate partition coefficients that are >1 and were therefore
depleted from the mantle relative to CI during core formation. They can be divided into three basic groups,
weakly siderophile (grey symbols), siderophile (unfilled symbols) and highly siderophile (black squares).
Lithophile elements are not depleted from the mantle as a result of core formation (all other black symbols).
An additional depletion trend affecting both lithophile and siderophile elements results from volatile
behavior (circles) which is considered to be a broad function of the elements’ 50% condensation
temperatures from the solar nebula at 10−4 bar (data from Wasson, 1985).
GG325 L33, F2013
Chemical indicators of core formation from the mantle
Before we interpret the data further, let's look at how we might
estimate the siderophile content of the early mantle.
Direct Method:
1. analyze actual samples of mantle (e.g., xenoliths in lava flow)
Indirect Methods:
1. analyze a siderophile element in a mantle-derived lava, and then
apply a melting model to estimate the element’s concentration in the
unmelted mantle source of that lava.
2. Analyze a “melting invariant” trace-element ratio of one siderophile
element and one non-siderophile, refractory element in a mantlederived lava, and then use some fancy footwork to estimate the
concentration of the refractory element using the chondritic primitive
mantle model.
GG325 L33, F2013
2
Chemical indicators of core formation from the mantle
Each method has assumptions and inherent weaknesses.
Direct # 1
Xenoliths are rare, and even rarer are xenoliths that haven’t either reacted
with their host magma or don’t show evidence of prior high-temperature
modification in the mantle (recrystallization, secondary phases, etc .
Indirect # 1
Melting models, as we’ll discuss in the near future, are good at predicting
relative abundances of trace elements in the source but are not as good at
predicting absolute concentrations.
Indirect # 2
For the most part, this is the most accurate and most widely used
approach because it depends on compositional characteristics that are
well known. However, assumptions about the constancy of a trace element
ratio during melting still result in some large uncertainties.
Let’s look at this method in more detail.
GG325 L33, F2013
Chemical indicators of core formation from the mantle
Indirect # 2
This method has 3 steps, outlined below.
Henry's Law, applied to equilibrium between a melt and one or
more solids, says that for an element A:
AK
d
=
[modal conc. of A]solids
[conc. of A]melt
Some siderophile elements have low Kd values (<1) during
melting of mantle rock to produce basalt, so that very little of
such elements will remain in the solid phases in equilibrium with
a basaltic melt.
GG325 L33, F2013
3
Chemical indicators of core formation from the mantle
Indirect # 2
(continued).
1. invariant ratios allow rely on analogous behavior to “see through” the
chemical effects of processes.
These involve elements that behave very similarly during a process, such as
melting of the mantle.
Non-siderophile example: K and U behave very similarly during mantle
melting, such that K/U  12,000 in many mantle-derived rocks on Earth
(i.e., K/U ratio is a nearly invariant ratio for most mantle melting systems).
 KKd (note: we used this invariance when we talked about formation
temperatures of planetary surfaces because K is more volatile than U)
UK
d
To calculate the siderophile content of the mantle, we want
invariant ratios that include a siderophile element. For example,
Mo/Nd. Mo is moderately siderophilic. Nd is a lithophile.
MoK  NdK
d
d
so (Mo/Nd)basalt  (Mo/Nd)mantle
GG325 L33, F2013
Chemical indicators of core formation from the mantle
Indirect # 2, continued.
2. Refractory elements give absolute concentration:
Nd is a refractory element, so it was not redistributed significantly during the
high temperature processes that occurred during accretion.
Al is another refractory element. Since it is a major constituent of the high
temperature materials that formed the Earth's mantle, the Al content of the
mantle is very well known.
Al content is a good thing to pin absolute concentration of our siderophile
element to (in this case, the element Mo).
It turns out that Ndvolatility Alvolatility, meaning they condense similarly during
the condensation sequence.
Therefore, it is safe to assume that
(Nd/Al)early mantle
(Nd/Al)chondrites
Remember, assuming that the early mantle formed from chondritic material
is a very good approximation for the refractory elements.
GG325 L33, F2013
4
Chemical indicators of core formation from the mantle
Indirect # 2, continued.
3. Absolute Siderophile Concentration Calculation
We can use the following equation to calculate the absolute concentration of
our siderophile element (Mo) in the mantle:
(Mo/Nd)old basalt x (Nd/Al)chondrites x Almantle = Momantle
Note that this final equation utilizes three well-known facts to determine a
fourth:
similar behavior of Mo and Nd during formation of a basalt from the mantle
similar volatility of Nd and Al during condensation and accretion.
the absolute concentration of Al in the mantle.
The same procedure can be applied to other siderophile elements too. For
instance, for W, one would use the invariant W/U ratio and the fact that
Uvolatility
Alvolatility to calculate Wmantle from Almantle
GG325 L33, F2013
Chemical indicators of core formation from the mantle
Some Results of Indirect # 2
Both Mo and W are considered "moderately" siderophile.
Applying the above procedure, we find that:
Momantle and W mantle are somewhere in between what would be
expected from complete equilibrium with molten Fe and no
equilibrium with molten Fe.
GG325 L33, F2013
5
Chemical indicators of core formation from the mantle
Some Results of Indirect # 2
On the other hand, the highly siderophile elements (e.g., platinum
group elements, denoted below as “PGE") appear to have
PGEmantle all in equilibrium with molten Fe
Predicted
PGEmantle
observed
And among equally siderophilic elements, the more-volatile ones
(“Y”) are enriched relative to the more-refractory ones (“X”) in the
mantle.
Xmantle all in equilibrium with molten Fe
Ymantle all in equilibrium with molten Fe
but Ymantle > Ymantle all in equilibrium with molten Fe
so that Yearly mantle > Xearly mantle
more-volatile
more-refractory
GG325 L33, F2013
Core Formation Chemistry Summary
Interpretation:
Some combination of 3 conditions governed core formation:
1) incomplete equilibrium of the whole mantle with molten Fe
2) some siderophile elements were added to the mantle after the
core formed
3) the molten Fe wasn't pure, thus changing the affinity of various
elements for the "polluted" molten Fe phase.
To distinguish 1 and 2, we compare the relative abundances of more
and less volatile elements of the same siderophilicity with the
relative abundances of same volatility and differing siderophilicity
elements.
To distinguish 1 and 3, we compare the relative abundances of
various siderophile elements between silicate and pure Fe versus
silicate with of impure Fe.
GG325 L33, F2013
6
For instance,
Recent experimental results
demonstrate the influence of S on the
exchange coefficients (KD) of V, Nb,
Mn, Ga, In, Zn, Cr and Ta into the
core.
They are displayed here as a function
of mol fraction of S in the metal (XS).
Clearly, some elements are sensitive
to S (such as V and Cr), whereas
others are not (such as In and Nb)
Filled symbols are results from this study
(2 GPa, 2023 K; 6 GPa, 2373 K; 9 GPa, 2373 K,
18 GPa, 2573 K); others from the literature.
Mann et al., 2009.
GG325 L33, F2013
Core Formation Chemistry and Homogeneous Accretion
This topic is very much an area of ongoing research, but it looks like
at least some volatile-rich chondritic material was added to the
mantle after the core was formed to explain the moderately
siderophile/very volatile element abundances.
This means that core formation occurred before accretion was
complete.
In general, the mantle can be considered to have been ~90%
equilibrated with pure molten Fe, which implies the Earth had built
up to at least 90% of it’s current size when core formation occurred.
Other evidence for this late chondritic veneer being added material
to Earth after core formation comes from estimates of Fe+3/Fe+2,
CO2/CO and H2O/H2 in the mantle.
All are slightly higher than expected for homogenous accretion, yet
lower than expected for heterogeneous accretion.
GG325 L33, F2013
7
Earth Accretion Summary
Both homogeneous and heterogeneous accretion fail to explain all observations,
but the homogeneous model is better at predicting the distribution of materials
within the early Earth.
Most geochemists now prefer a modified homogeneous accretion model (i.e.,
mostly homogeneous but with some funny business near the end).
1. Homogeneous accretion of perhaps 90% of the present mass.
2. Major part of the core forms, stripping highly siderophile elements from the
mantle. Moderately siderophile elements (e.g., Mo and W) are not removed as
effectively (possibly for kinetic reasons or because molten Fe is polluted by
FeS, FeO or another contaminant; we don’t know yet).
3. Most of the last ~10% of the Earth is accreted and the last few percent of
the core forms, during which some additional stripping of siderophile elements
(including moderately siderophilic ones) occurs.
4. A veneer of chondritic material is added, increasing all siderophile element
concentrations, the relative proportion of volatile siderophiles to non-volatile
siderophiles, and the abundances of other volatiles in the mantle.
As we’ll see, this model requires further modification to account for our atmosphere.
GG325 L33, F2013
The formation of our Moon
Earth’s moon plays a role in the story of our own accretion and
differentiation.
The moon is:
1. of relatively large mass for a planetary satellite
2. in an odd high inclination orbit
3. with Earth, retains a great deal of angular momentum
4. is much more dense than most moons and unique of all
satellites in being essentially without “ice” components
5. compositionally very similar to Earth in some aspects.
GG325 L33, F2013
8
GG325 L33, F2013
The formation of our Moon
Limited seismic data from
moonquakes indicate an
internal layering broadly similar
to the Earth’s.
The moon has a thick crust and
an upper and lower mantle, but
only a tiny core (2-4% of the
Moon’s mass vs. 32.5% for
Earth’s core).
The small lunar core may be
some combination of Fe metal
and Fe sulfide. It’s not clear
whether it’s partially molten or
not. The moon has only a very
weak magnetic field.
GG325 L33, F2013
9
The formation of our Moon
Lunar sampling expeditions
provide invaluable samples
of the lunar crust.
GG325 L33, F2013
The formation of our Moon
The moon is compositionally very similar to Earth in some
aspects.
oxygen isotopic composition
bulk major element composition,
but in the moon’s case there’s much less of:
metallic Fe (overall), although the silicate portion of the
moon has more Fe2+ than the silicate Earth
siderophile elements (e.g., Ga, Ge)
moderately volatile elements (e.g., Na, K, Rb, Cs)
highly volatile elements (e.g., Bi, Pb, As).
And there’s significantly more of
refractory elements Al, Ca, Ti
GG325 L33, F2013
10
GG325 L33, F2013
The formation of our Moon
Of the four likely scenarios for
the formation of the moon:
capture of an extra terrestrial
body.
”Auto” fission from the Earth
Separate condensation near
Earth
”Forced” fission from Earth,
via impact (“collision”)
the latter is the most commonly
accepted, because it explains
most of the observed facts.
GG325 L33, F2013
11
The formation of our Moon
The giant impact scenario is supported by:
 the relative proportions of refractory and volatile
elements on the Earth and Moon
 the E-M homogeneity of refractory-element ratios
 and similar E-M oxygen-isotope ratios
Abundances of refractory and volatile elements in the Earth and Moon
Earth
Moon
Moon/Earth
K ppm
150-200
100
0.50-0.67
U ppm
15-20
40
2.0-2.7
K/U
10,000
2500
0.25
Th ppb
57-76
152
2.0-2.7
Rb ppm
0.50-0.66
0.30
0.45-0.6
Sr ppm
16-21
40
1.9-2.5
Rb/Sr
0.031
0.0075
Na ppm
1200-1600
800
-0.50-0.67
Source: Taylor, 1979
GG325 L33, F2013
Formation of our moon
The giant impact hypothesis: First proposed by R. Daly in the 1940s,
and later by Hartmann and Davis (1974).
Current theory is that the Moon formed when the Earth (at >0.5 its
present size) was struck at low angle by a slow-moving (~5 km/s) body
slightly larger than Mars during the latest stages of accretion.
The impact would have occurred after core formation on both the Earth
and the impacting body, which is sometimes called “Theia.”
Theia is postulated to have accreted from material in roughly the same
annulus of the solar nebula as the Earth (supported by the O-isotope
data) and presumably was similarly
depleted in highly and moderately volatile
elements relative to the chondrites.
The collision partially disrupted/melted
the Earth’s mantle and completely
disrupted/melted Theia.
GG325 L33, F2013
12
Theia Impact Simulation
Simulation of the early
stages of the Moon’s
origin by a giant impact.
(Canup, 2004).
GG325 L33, F2013
Formation of The Moon
Much of the resulting impact
debris went into Earth orbit.
High temperatures after the
impact evaporated the most
volatile elements, leaving the
moon very depleted in such
material.
Most of the core of the Theia is believed to have accreted to the Earth
and the orbiting silicate material slowly coalesced to form the Moon.
About 85% of the material forming the moon was derived from Theia
and about 15% was from Earth’s mantle.
The lunar siderophile element depletion is explained by prior core
formation from Theia and Earth, and generation of the moon from the
silicate fractions. A small core then segregated from the largely
molten moon.
GG325 L33, F2013
13
Formation of our moon
Lunar rocks indicate very extensive
melting occurred just after formation,
probably producing a magma ocean >100
km deep (maybe more).
Melting produced a basaltic crust that was
largely emplaced within 100-200 Myr after
the Moon’s formation. The very oldest
lunar rocks yet dated are nearly 4.5 Ga.
Heavy impact bombardment continued
until ~3.9 Ga.
From about 3.8-3.1 Ga, the large impact
craters were flooded with immense
outpourings of tholeitic basalt, after which
volcanic activity ceased.
GG325 L33, F2013
Formation of our moon
The Moon’s basaltic crust contains a range of rock types. Two extreme
compositions are found:
(a) a variety of basalt that is enriched in K, the lanthanide rare-earths, and
phosphorus; this rock type is termed “KREEP”.
(b) anorthosite (anorthite-rich rock) dominates the lunar highlands; it’s
thought to be produced by
plagioclase that floated to the
top of the magma ocean.
These two rock types form
the upper and lower
extremes of mirror-image
normalized rare-earth
element profiles, suggesting
that they separated from
the same homogeneous
reservoir.
GG325 L33, F2013
14
Formation of the Moon
Other implications of lunar formation:
The total amount of energy released by accretion of the Earth was much
greater than that released by the collision of Theia. Thus we expect that,
like the Moon, the early Earth melted extensively.
This should have produced an early proto-crust rather similar to that on the
Moon, and a partially depleted mantle.
http://maps.jpl.nasa.gov/textures/
GG325 L33, F2013
Terrestrial Planet Formation Summary
•
•
•
•
•
•
Rapid collapse of molecular gas
cloud; condensation; high-energy
processing.
Small (km size) bodies form quickly
(<106 yr). Some of these bodies
differentiate (for impact and
radiodecay heating, i.e., of 26Al,
60Fe).
Moon- and Mars-size bodies may
also form nearly as quickly.
Slower build up of larger bodies:
time ~107 - 108 yr.
Last stages of accretion occur in the
absence of the solar nebula (i.e.,
after T Tauri stage).
Rapid formation of
kilometer bodies from dust
Rapid Formation of Moon
sized bodies by runaway
accretion
Slow (~10 Ma) Formation of
Earthlike Planets
Mixing across ~1AU or more likely
(with chemical disequilibrium?)
during later stages of accretion.
GG325 L33, F2013
15